Nabble - pacman-dev

Re: makepkg in a constrained environment

2009-12-11T08:38:29Z

On 03/12/2009, at 8:41 AM, Dan McGee wrote:

> On Wed, Dec 2, 2009 at 2:47 PM, Ciprian Dorin, Craciun
> <ciprian.craciun@...> wrote:
>> Looking closer at what the mentioned patch introduced:
>> ~~~~
>> - local sum="$(${integ}sum "$file" | cut -d ' ' -f 1)"
>> + local sum="$(openssl dgst -${integ} "$file" | awk '{print $NF}')"
>> ~~~~
>> I'm proposing not to get back to what it has been (which assumed
>> that there was a md5sum sha128sum, <something-else>sum), but I'm
>> proposing that for md5 we should use md5sum, which I'm sure it exists
>> on any UNIX. Or better even, try to see if openssl exists, if not
>> falling back to md5sum, at least for md5sums...
>
> As a side note, it doesn't exist on all UNIX-y systems, for example on
> Mac OS X- there is a /sbin/md5, but not an md5sum command.
>
> -Dan
>

From what I recall, the `md5` utility on Mac OS X uses openssl
(libcrypto) anyway. It does occur to me that hashing should be re-
factored into a separate function, as it does occur at least twice.

On 03/12/2009, at 5:30 AM, Ciprian Dorin, Craciun wrote:

> For the moment, I think that my best option is to write my own
> makepkg.
>
> Of course I'll write it in any other programming language than
> Bash... :) For example I think I'll try rc (Plan9 shell), because it
> seems the most sane shell I've ever seen...

That's a little drastic. Why not just make the changes you wanted to
makepkg? You could also send the patches to the ML. They might not get
accepted, but if you're willing to go as rewriting makepkg from
scratch, what do you care?

Bash isn't the greatest language to write a large (for bash) utility
like makepkg, but I doubt RC will be much better.

Why can't you just install bsdtar and openssl? Sure they're separate
packages, but ones that makepkg depends on.

Re: [PATCH] makepkg: improve removal of installed dependencies

2009-12-09T21:39:58Z

Cedric Staniewski wrote:

> On 12/03/2009 03:33 PM, Allan McRae wrote:
>> Compare a list of packages on the system before and after dependency resolution
>> in order to get a complete list of packages to remove. This allows makepkg to
>> remove packages installed due to provides.
>>
>> Bail in cases where packages that were on the system originally have been
>> removed as there is a risk of breaking the system when removing the new
>> packages
>>
>> Fixes FS#15144
>>
>> Signed-off-by: Allan McRae <allan@...>
>> ---
>
> I have not tested the patch yet, so just some comments on the code for now.
>

Thanks for the comments. I have pushed the updated patch on my working
branch:
http://projects.archlinux.org/users/allan/pacman.git/commit/?h=working&id=ffc7b0dc

<snip>

>
> You could use comm here. The advantage would be that it do not pull in
> a new dependency as it is in coreutils too and you do not need grep to
> get the unique lines.
>
> if [[ -n $(printf "%s\n" ${original_pkglist[@]} \
> | comm -23 - <(printf "%s\n" ${current_pkglist[@]})) ]]
>

I went for:
if [[ -n $(comm -23 <(printf "%s\n" "${original_pkglist[@]}") \
<(printf "%s\n" "${current_pkglist[@]}")) ]]; then

The only slight annoyance was when using comm the output of pacman -Qq
needed sorted as pacman outputs in a very slightly different
alphabetical order to what comm expects and "--nocheck-order" is not
portable to BSD/OSX.

Allan

Re: [PATCH 2/2] Add new --print operation for all operations

2009-12-09T16:22:14Z

On Mon, Nov 16, 2009 at 4:02 AM, Dan McGee <dpmcgee@...> wrote:

>> +*\--print*::
>> + Only print the targets instead of performing the actual operation (sync,
>> + remove or upgrade). Use '\--print-format' to specify how targets are
>> + displayed.
>> +
>> +*\--print-format* <'format'>::
>> + Specify a printf-like format to control the output of the '\--print'
>> + operation. The possible are attributes are : %n for pkgname, %v for pkgver, %l
>> + for location, %r for repo and %s for size. The default format string is "%l",
>> + which displays url with '-S', filename with '-U' and pkgname-pkgver with '-R'.
>> +
>
> * Isn't --print actually -p/--print?
> * Should the default format stuff actually be in the --print
> documentation (either as well, or completely move it there?). I feel
> like I would look there first.
> * You didn't kill the print-uri documentation.
>

done done done

>> + /* set up the print operations */
>> + if(config->print) {
>> + config->noconfirm = 1;
>> + config->flags |= PM_TRANS_FLAG_NOCONFLICTS;
>> + config->flags |= PM_TRANS_FLAG_NOLOCK;
>> + /* Display only errors */
>> + config->logmask &= ~PM_LOG_WARNING;
> Man, I hate having to do this. We really should think about a
> stderr/stdout jihad.
>

So like we would display all warnings/errors to stderr and the useful
stuff to stdout ?

And it indeed sounds like it would be a lot of work.. It seems like
pacman was not written having in mind it could be used in
scripts/wrapper.

>> + }
>> +
>> +
>> #if defined(HAVE_GETEUID) && !defined(CYGWIN)
>> /* check if we have sufficient permission for the requested operation */
>> if(myuid > 0 && needs_root()) {
>> diff --git a/src/pacman/remove.c b/src/pacman/remove.c
>> index b5119fa..06503d3 100644
>> --- a/src/pacman/remove.c
>> +++ b/src/pacman/remove.c
>> @@ -159,6 +159,12 @@ int pacman_remove(alpm_list_t *targets)
>> }
>>
>> /* Step 3: actually perform the removal */
>> + if(config->print) { /* --print */
>> + print_packages(alpm_trans_get_pkgs());
>> + /* we are done */
> None of the comments here seem particularly necessary. (/*--print*/,
> /*we are done*/). Can you kill them?
>

done

>> - /* Display only errors with -Sp and -Sw operations */
>> - if((config->flags & PM_TRANS_FLAG_DOWNLOADONLY) || config->op_s_printuris) {
>> + /* Display only errors with -Sw operations */
>> + if(config->flags & PM_TRANS_FLAG_DOWNLOADONLY) {
> Why do we do this for -Sw?
>

That question was not directly related to my patch but yeah, I have no
idea why we do this for -Sw so I killed it.

>>
>> /* Step 3: perform the installation */
>> + if(config->print) { /* --print */
>> + print_packages(alpm_trans_get_pkgs());
>> + /* we are done */
>> + trans_release();
>> + return(0);
> Kill comments

done

>> + /* %r : repo */
>> + if(strstr(temp,"%r")) {
>> + const char *repo = "local";
>> + pmdb_t *db = alpm_pkg_get_db(pkg);
>> + if(db) {
>> + repo = alpm_db_get_name(db);
>> + }
>> + string = strreplace(temp, "%r", repo);
>> + free(temp);
>> + temp = string;
>> + }
>> + /* %s : size */
>> + if(strstr(temp,"%s")) {
>> + char *size;
>> + double mbsize = 0.0;
>> + mbsize = pkg_get_size(pkg) / (1024.0 * 1024.0);
>> + asprintf(&size, "%.2f", mbsize);
>> + string = strreplace(temp, "%s", size);
>> + free(size);
>> + free(temp);
>> + temp = string;
>> + }
> This strreplace() stuff seems a bit naive, considering we could do a
> lot of replaces on the string. I'm not sure we need to worry about it
> but could be a place for some improvement.
>

Do you mean that we are walking through the same string many times ?

I thought at some point we could have a multireplace function which
could take a list of struct replace { char *old ; char *new; } but ...
firstly it looked a bit overkill
and secondly, I was worried that all the pkg/db accessors could be
more costly than these strings operations.
Though that could also be solved by doing a first pass on the string
to see what format options we have, and then build the replace list
with only the stuff we need.

I agree the current code is quite naive but it does not affect
performance so much that it makes the operation unusable.
pacman -S kde gnome --print --printformat '%n %v %l %r %s' takes 0.7 second.

>
> OK, and now for some usability testing.
>
> $ ./src/pacman/pacman -Q
> <big list of packages>
> ...
> zope-interface 3.5.1-1
> zvbi 0.2.33-1
>
> $ ./src/pacman/pacman -Q --print
> error: no targets specified (use -h for help)
>
> So two things here- I expected --print to work and it did not, and we
> already have the standard "pkgname <space> pkgver" output that our
> default print format uses. Do we want to invent a new one with the
> default --print for remove?
>
> Oh wait, does --print even work for -Q? Now I'm all confused. :)
> $ ./src/pacman/pacman -Q --print pacman-git
> error: package "pacman-git" not found
>
> $ ./src/pacman/pacman -Q pacman-git
> pacman-git 20091018-1
>
> More unexpected behavior...
>
> $ ./src/pacman/pacman -Ss --print pacman
> core/pacman 3.3.3-1 (base)
> A library-based package manager with dependency support
> core/pacman-mirrorlist 20090616-1 (base) [installed]
> Arch Linux mirror list for use by pacman
> extra/namcap 2.4-1 [installed]
> A Pacman package analyzer
>
> I wanted a list of URLs for all packages that matched my pacman
> search, I did not get that...
>
> -Dan
>
>

About that last part, I already replied one month ago :)
Since the changes I just did were quite trivial, I will just push an
updated version to my print branch.

[PATCH] improve download_internal error messages

2009-12-09T14:30:11Z

download_internal is supposed to always set pm_errno but did not in many
cases.

The most important (and tested) change is the one concerning fetchStat. This
is typically where the code will fail when the network is down for example.

Before commit d2dbb04a9af7a18da, this fetchStat call did not exist and the
same kind of errors would be encountered in the fetchXGet call that follows.
I just copied the error printing to restore the old behavior.

Signed-off-by: Xavier Chantry <shiningxc@...>
---
lib/libalpm/dload.c | 10 ++++++++--
1 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/lib/libalpm/dload.c b/lib/libalpm/dload.c
index bc5cdfd..499f7f5 100644
--- a/lib/libalpm/dload.c
+++ b/lib/libalpm/dload.c
@@ -118,13 +118,14 @@ static int download_internal(const char *url, const char *localpath,

filename = get_filename(url);
if(!filename) {
- return(-1);
+ _alpm_log(PM_LOG_ERROR, _("url '%s' is invalid\n"), url);
+ RET_ERR(PM_ERR_SERVER_BAD_URL, -1);
}

fileurl = fetchParseURL(url);
if(!fileurl) {
_alpm_log(PM_LOG_ERROR, _("url '%s' is invalid\n"), url);
- RET_ERR(PM_ERR_SERVER_BAD_URL, -1);
+ RET_ERR(PM_ERR_LIBFETCH, -1);
}

destfile = get_destfile(localpath, filename);
@@ -179,6 +180,9 @@ static int download_internal(const char *url, const char *localpath,
* non-stat request, so avoid it. */
fetchLastErrCode = 0;
if(fetchStat(fileurl, &ust, "") == -1) {
+ pm_errno = PM_ERR_LIBFETCH;
+ _alpm_log(PM_LOG_ERROR, _("failed retrieving file '%s' from %s : %s\n"),
+ filename, gethost(fileurl), fetchLastErrString);
ret = -1;
goto cleanup;
}
@@ -230,6 +234,7 @@ static int download_internal(const char *url, const char *localpath,
dl_thisfile = 0;
localf = fopen(tempfile, "wb");
if(localf == NULL) { /* still null? */
+ pm_errno = PM_ERR_RETRIEVE;
_alpm_log(PM_LOG_ERROR, _("error writing to file '%s': %s\n"),
tempfile, strerror(errno));
ret = -1;
@@ -247,6 +252,7 @@ static int download_internal(const char *url, const char *localpath,
size_t nwritten = 0;
nwritten = fwrite(buffer, 1, nread, localf);
if((nwritten != nread) || ferror(localf)) {
+ pm_errno = PM_ERR_RETRIEVE;
_alpm_log(PM_LOG_ERROR, _("error writing to file '%s': %s\n"),
tempfile, strerror(errno));
ret = -1;
--
1.6.5.5

Re: pacman -Sy with no network = weird result

2009-12-09T14:19:14Z

On Mon, Nov 30, 2009 at 7:03 AM, Dan McGee <dpmcgee@...> wrote:

> On Thu, Nov 26, 2009 at 9:52 AM, Xavier <shiningxc@...> wrote:
>> Here is a quick and lazy bug report so that I do not forget :)
>>
>> LANG=C sudo pacman -Sy
>> :: Synchronizing package databases...
>> error: failed to update testing (library not initialized)
>> error: failed to update core (library not initialized)
>> error: failed to update extra (library not initialized)
>> error: failed to update community-testing (library not initialized)
>> error: failed to update community (library not initialized)
>> error: failed to synchronize any databases
>>
>> The error and the output is quite confusing.
>
> Should this get turned into a real bug report? :P
>

No, it should get turned into a patch :)

I just realized an additional reason I got very confused by the above
output (I thought my system was completely broken :D). I think I was
used to the old ouput :
:: Synchronizing package databases...
error: failed retrieving file 'testing.db.tar.gz' from
mir.archlinux.fr : Transient resolver failure
error: failed retrieving file 'testing.db.tar.gz' from
mir1.archlinuxfr.org : Transient resolver failure
error: failed to update testing (Transient resolver failure)
error: failed retrieving file 'core.db.tar.gz' from mir.archlinux.fr :
Transient resolver failure
error: failed retrieving file 'core.db.tar.gz' from
mir1.archlinuxfr.org : Transient resolver failure
error: failed to update core (Transient resolver failure)
error: failed retrieving file 'extra.db.tar.gz' from mir.archlinux.fr
: Transient resolver failure
error: failed retrieving file 'extra.db.tar.gz' from
mir1.archlinuxfr.org : Transient resolver failure
error: failed to update extra (Transient resolver failure)
error: failed retrieving file 'community.db.tar.gz' from
mir.archlinux.fr : Transient resolver failure
error: failed retrieving file 'community.db.tar.gz' from
mir1.archlinuxfr.org : Transient resolver failure
error: failed to update community (Transient resolver failure)
error: failed to synchronize any databases

So not only we did not set pm_errno (to ERR_LIBFETCH which prints
fetchLastErrString = Transient resolver failure), but also we no
longer printed the message for each url/mirror.

This was caused by commit d2dbb04a9af7a18da which does fetchStat first
, but did not print any error in that case.
+ fetchLastErrCode = 0;
+ if(fetchStat(fileurl, &ust, "") == -1) {
+ ret = -1;
+ goto cleanup;
+ }

A patch will follow to restore the old more verbose behavior :)

Sidenote : I saw many times people confused, even with the above
verbose output, usually after installing arch and before setting up
the network. Not sure how they cannot realize their network is down.
Maybe 'transient resolver failure' is confusing ? That's a libfetch
message though, so we cannot change it. And it seems appropriate.
And 'failed retrieving file' sounds explicit enough.

Comments and thoughts welcome.

Re: [PATCH] docs: set text width to 72 in vim directive

2009-12-09T10:50:15Z

On Thu, Dec 3, 2009 at 8:59 AM, Cedric Staniewski <cedric@...> wrote:

> On 12/03/2009 01:47 AM, Dan McGee wrote:
>> On Wed, Dec 2, 2009 at 12:05 PM, Cedric Staniewski <cedric@...> wrote:
>>
>> A "why" in the commit message would be awfully nice, I'm confused here. :)
>>
>>> Signed-off-by: Cedric Staniewski <cedric@...>
>
> Sorry, I thought it were quite clear. The text in the man pages is
> already wrapped after a certain amount of characters and I thought is
> would be nice to add this to the vim directive or am I missing something
> here?
>
>

These man pages wrapping have always pissed me off , because all lines
are indented by a tab and vim does not handle that.
When I do 'gq' on one text block, it only keeps the tab on the first
line... So I always had to manually re-insert the tab for all the next
lines. And the result is that the first line is always shorter than
all the following ones (by one tab).

What am I missing ?

Re: [PATCH] makepkg: remove srclinks directory on error exit

2009-12-09T10:41:57Z

On Thu, Dec 3, 2009 at 10:43 AM, Allan McRae <allan@...> wrote:
>
> OK, your comment about the external commands being used having good error
> messages has me convinced something needs done. It just seems a strange
> place for this particular clean up. Saying that, I'm not sure what would be
> a better place unless we had an error trap just for that function which
> passed on to trap_exit.
>
> Dan, Xavier, others: any comments?
>

It might look a bit strange, but I am fine with that patch anyway.

Re: [PATCH] makepkg: improve removal of installed dependencies

2009-12-07T06:54:20Z

On 12/03/2009 03:33 PM, Allan McRae wrote:

> Compare a list of packages on the system before and after dependency resolution
> in order to get a complete list of packages to remove. This allows makepkg to
> remove packages installed due to provides.
>
> Bail in cases where packages that were on the system originally have been
> removed as there is a risk of breaking the system when removing the new
> packages
>
> Fixes FS#15144
>
> Signed-off-by: Allan McRae <allan@...>
> ---

I have not tested the patch yet, so just some comments on the code for now.

>
> This moves the generation of the installed package list to directly after
> the installation of deps to prevent issues if the user installs packages
> while the build is occurring:
> http://bugs.archlinux.org/task/15144#comment53938
>
> Rebased on Cedric's run_pacman and $PACMAN patches.
>
> doc/makepkg.8.txt | 6 +++---
> scripts/makepkg.sh.in | 39 +++++++++++++++++++++------------------
> 2 files changed, 24 insertions(+), 21 deletions(-)
>
> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
> index ad27bca..168b369 100644
> --- a/doc/makepkg.8.txt
> +++ b/doc/makepkg.8.txt
> @@ -166,9 +166,9 @@ Environment Variables
> ---------------------
> *PACMAN*::
> The command that will be used to check for missing dependencies and to
> - install and remove packages. Pacman's -U, -T, -S and -Rns operations
> - must be supported by this command. If the variable is not set or
> - empty, makepkg will fall back to `pacman'.
> + install and remove packages. Pacman's -Qq, -Rns, -S, -T, and -U
> + operations must be supported by this command. If the variable is not
> + set or empty, makepkg will fall back to `pacman'.
>
>
> Additional Features
> diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in
> index c730150..5743ea4 100644
> --- a/scripts/makepkg.sh.in
> +++ b/scripts/makepkg.sh.in
> @@ -27,8 +27,10 @@
>
> # makepkg uses quite a few external programs during its execution. You
> # need to have at least the following installed for makepkg to function:
> -# bsdtar (libarchive), bzip2, coreutils, fakeroot, find (findutils),
> -# getopt (util-linux), gettext, grep, gzip, openssl, sed, tput (ncurses)
> +# bsdtar (libarchive), bzip2, coreutils, diff (diffutils), fakeroot,
> +# find (findutils), getopt (util-linux), gettext, grep, gzip, openssl,
> +# sed, tput (ncurses)
> +
>
> # gettext initialization
> export TEXTDOMAIN='pacman'
> @@ -343,7 +345,7 @@ download_file() {
>
> run_pacman() {
> local ret=0
> - if (( ! ASROOT )) && [[ $1 != "-T" ]] && sudo -l $PACMAN &>/dev/null; then
> + if (( ! ASROOT )) && [[ $1 != "-T" && $1 != "-Qq" ]] && sudo -l $PACMAN &>/dev/null; then
> sudo $PACMAN $PACMAN_OPTS "$@" || ret=$?
> else
> $PACMAN $PACMAN_OPTS "$@" || ret=$?
> @@ -398,7 +400,6 @@ handle_deps() {
> }
>
> resolve_deps() {
> - # $pkgdeps is a GLOBAL variable, used by remove_deps()
> local R_DEPS_SATISFIED=0
> local R_DEPS_MISSING=1
>
> @@ -408,7 +409,6 @@ resolve_deps() {
> fi
>
> if handle_deps $deplist; then
> - pkgdeps="$pkgdeps $deplist"
> # check deps again to make sure they were resolved
> deplist="$(check_deps $*)"
> [[ -z $deplist ]] && return $R_DEPS_SATISFIED
> @@ -425,23 +425,25 @@ resolve_deps() {
> return $R_DEPS_MISSING
> }
>
> -# fix flyspray bug #5923
> remove_deps() {
> - # $pkgdeps is a GLOBAL variable, set by resolve_deps()
> (( ! RMDEPS )) && return
> - [[ -z $pkgdeps ]] && return
>
> - local dep depstrip deplist
> - deplist=""
> - for dep in $pkgdeps; do
> - depstrip="${dep%%[<=>]*}"
> - deplist="$deplist $depstrip"
> - done
> + # check for packages removed during dependency install (e.g. due to conflicts)
> + # removing all installed packages is risky in this case
> + if (( $(diff <(printf "%s\n" "${original_pkglist[@]}") \
> + <(printf "%s\n" "${current_pkglist[@]}") | grep "<" | wc -l) > 0 )); then
> + warning "$(gettext "Failed to remove installed dependencies.")"
> + return 0
> + fi
>

You could use comm here. The advantage would be that it do not pull in
a new dependency as it is in coreutils too and you do not need grep to
get the unique lines.

if [[ -n $(printf "%s\n" ${original_pkglist[@]} \
| comm -23 - <(printf "%s\n" ${current_pkglist[@]})) ]]

Note the code above is untested.

> - msg "Removing installed dependencies..."
> + local deplist=($(diff <(printf "%s\n" "${original_pkglist[@]}") \
> + <(printf "%s\n" "${current_pkglist[@]}") | grep ">"))
> + deplist=(${deplist[@]#>})

local deplist=($(printf "%s\n" ${original_pkglist[@]} \
| comm -13 - <(printf "%s\n" ${current_pkglist[@]})))

> + [ ${#deplist[@]} -eq 0 ] && return

(( ${#deplist[@]} == 0 )) :)

>
> + msg "Removing installed dependencies..."
> # exit cleanly on failure to remove deps as package has been built successfully
> - if ! run_pacman -Rns $deplist; then
> + if ! run_pacman -Rn ${deplist[@]}; then
> warning "$(gettext "Failed to remove installed dependencies.")"
> return 0
> fi
> @@ -1874,14 +1876,13 @@ if (( SOURCEONLY )); then
> exit 0
> fi
>
> -# fix flyspray bug #5973
> if (( NODEPS || NOBUILD || REPKG )); then
> # no warning message needed for nobuild, repkg
> if (( NODEPS )); then
> warning "$(gettext "Skipping dependency checks.")"
> fi
> elif [ $(type -p "${PACMAN%% *}") ]; then
> - unset pkgdeps # Set by resolve_deps() and used by remove_deps()
> + original_pkglist=($(run_pacman -Qq)) # required by remove_deps

Could be skipped if RMDEPS == 0 but I do not know how fast / slow this
pacman operation actually is.

> deperr=0
>
> msg "$(gettext "Checking Runtime Dependencies...")"
> @@ -1890,6 +1891,8 @@ elif [ $(type -p "${PACMAN%% *}") ]; then
> msg "$(gettext "Checking Buildtime Dependencies...")"
> resolve_deps ${makedepends[@]} || deperr=1
>
> + current_pkglist=($(run_pacman -Qq)) # required by remove_deps
> +

Same as above.

> if (( deperr )); then
> error "$(gettext "Could not resolve all dependencies.")"
> exit 1

[PATCH 3/3] add some more sanity check for optarg

2009-12-05T14:49:07Z

getopt should already ensure that optarg is not NULL when an argument is
required, but just be extra safe and double check it before using optarg.

To be honest, I only did that to make clang shut up and eliminate the last
warnings it reported.

Signed-off-by: Xavier Chantry <shiningxc@...>
---
src/pacman/pacman.c | 16 +++++++++++++++-
1 files changed, 15 insertions(+), 1 deletions(-)

diff --git a/src/pacman/pacman.c b/src/pacman/pacman.c
index 38fc560..522da1b 100644
--- a/src/pacman/pacman.c
+++ b/src/pacman/pacman.c
@@ -334,6 +334,8 @@ static void setlibpaths(void)
}
}

+#define check_optarg() if(!optarg) { return(1); }
+
/** Parse command-line arguments for each operation.
* @param argc argc
* @param argv argv
@@ -408,12 +410,14 @@ static int parseargs(int argc, char *argv[])
case 0: break;
case OP_NOCONFIRM: config->noconfirm = 1; break;
case OP_CONFIG:
+ check_optarg();
if(config->configfile) {
free(config->configfile);
}
config->configfile = strndup(optarg, PATH_MAX);
break;
case OP_IGNORE:
+ check_optarg();
list = strsplit(optarg, ',');
for(item = list; item; item = alpm_list_next(item)) {
alpm_option_add_ignorepkg((char *)alpm_list_getdata(item));
@@ -445,8 +449,13 @@ static int parseargs(int argc, char *argv[])
break;
case OP_NOPROGRESSBAR: config->noprogressbar = 1; break;
case OP_NOSCRIPTLET: config->flags |= PM_TRANS_FLAG_NOSCRIPTLET; break;
- case OP_ASK: config->noask = 1; config->ask = atoi(optarg); break;
+ case OP_ASK:
+ check_optarg();
+ config->noask = 1;
+ config->ask = atoi(optarg);
+ break;
case OP_CACHEDIR:
+ check_optarg();
if(alpm_option_add_cachedir(optarg) != 0) {
pm_printf(PM_LOG_ERROR, _("problem adding cachedir '%s' (%s)\n"),
optarg, alpm_strerrorlast());
@@ -457,9 +466,11 @@ static int parseargs(int argc, char *argv[])
config->flags |= PM_TRANS_FLAG_ALLDEPS;
break;
case OP_LOGFILE:
+ check_optarg();
config->logfile = strndup(optarg, PATH_MAX);
break;
case OP_IGNOREGROUP:
+ check_optarg();
list = strsplit(optarg, ',');
for(item = list; item; item = alpm_list_next(item)) {
alpm_option_add_ignoregrp((char *)alpm_list_getdata(item));
@@ -471,6 +482,7 @@ static int parseargs(int argc, char *argv[])
config->flags |= PM_TRANS_FLAG_ALLEXPLICIT;
break;
case OP_ARCH:
+ check_optarg();
setarch(optarg);
break;
case 'Q': config->op = (config->op != PM_OP_MAIN ? 0 : PM_OP_QUERY); break;
@@ -480,6 +492,7 @@ static int parseargs(int argc, char *argv[])
case 'U': config->op = (config->op != PM_OP_MAIN ? 0 : PM_OP_UPGRADE); break;
case 'V': config->version = 1; break;
case 'b':
+ check_optarg();
config->dbpath = strdup(optarg);
break;
case 'c':
@@ -516,6 +529,7 @@ static int parseargs(int argc, char *argv[])
config->quiet = 1;
break;
case 'r':
+ check_optarg();
config->rootdir = strdup(optarg);
break;
case 's':
--
1.6.5.4

[PATCH 2/3] refactor _parseoptions

2009-12-05T14:49:06Z

This function was quite huge (~230 lines) and difficult to parse, now it is
slightly better.

Signed-off-by: Xavier Chantry <shiningxc@...>
---
src/pacman/pacman.c | 308 ++++++++++++++++++++++++++++-----------------------
1 files changed, 170 insertions(+), 138 deletions(-)

diff --git a/src/pacman/pacman.c b/src/pacman/pacman.c
index 3dccd48..38fc560 100644
--- a/src/pacman/pacman.c
+++ b/src/pacman/pacman.c
@@ -713,6 +713,127 @@ cleanup:
return(ret);
}

+static int _parse_options(char *key, char *value)
+{
+ if(value == NULL) {
+ /* options without settings */
+ if(strcmp(key, "UseSyslog") == 0) {
+ alpm_option_set_usesyslog(1);
+ pm_printf(PM_LOG_DEBUG, "config: usesyslog\n");
+ } else if(strcmp(key, "ILoveCandy") == 0) {
+ config->chomp = 1;
+ pm_printf(PM_LOG_DEBUG, "config: chomp\n");
+ } else if(strcmp(key, "ShowSize") == 0) {
+ config->showsize = 1;
+ pm_printf(PM_LOG_DEBUG, "config: showsize\n");
+ } else if(strcmp(key, "UseDelta") == 0) {
+ alpm_option_set_usedelta(1);
+ pm_printf(PM_LOG_DEBUG, "config: usedelta\n");
+ } else if(strcmp(key, "TotalDownload") == 0) {
+ config->totaldownload = 1;
+ pm_printf(PM_LOG_DEBUG, "config: totaldownload\n");
+ } else {
+ pm_printf(PM_LOG_ERROR, _("directive '%s' without value not recognized\n"), key);
+ return(1);
+ }
+ } else {
+ /* options with settings */
+ if(strcmp(key, "NoUpgrade") == 0) {
+ setrepeatingoption(value, "NoUpgrade", alpm_option_add_noupgrade);
+ } else if(strcmp(key, "NoExtract") == 0) {
+ setrepeatingoption(value, "NoExtract", alpm_option_add_noextract);
+ } else if(strcmp(key, "IgnorePkg") == 0) {
+ setrepeatingoption(value, "IgnorePkg", alpm_option_add_ignorepkg);
+ } else if(strcmp(key, "IgnoreGroup") == 0) {
+ setrepeatingoption(value, "IgnoreGroup", alpm_option_add_ignoregrp);
+ } else if(strcmp(key, "HoldPkg") == 0) {
+ setrepeatingoption(value, "HoldPkg", option_add_holdpkg);
+ } else if(strcmp(key, "SyncFirst") == 0) {
+ setrepeatingoption(value, "SyncFirst", option_add_syncfirst);
+ } else if(strcmp(key, "Architecture") == 0) {
+ if(!alpm_option_get_arch()) {
+ setarch(value);
+ }
+ } else if(strcmp(key, "DBPath") == 0) {
+ /* don't overwrite a path specified on the command line */
+ if(!config->dbpath) {
+ config->dbpath = strdup(value);
+ pm_printf(PM_LOG_DEBUG, "config: dbpath: %s\n", value);
+ }
+ } else if(strcmp(key, "CacheDir") == 0) {
+ if(alpm_option_add_cachedir(value) != 0) {
+ pm_printf(PM_LOG_ERROR, _("problem adding cachedir '%s' (%s)\n"),
+ value, alpm_strerrorlast());
+ return(1);
+ }
+ pm_printf(PM_LOG_DEBUG, "config: cachedir: %s\n", value);
+ } else if(strcmp(key, "RootDir") == 0) {
+ /* don't overwrite a path specified on the command line */
+ if(!config->rootdir) {
+ config->rootdir = strdup(value);
+ pm_printf(PM_LOG_DEBUG, "config: rootdir: %s\n", value);
+ }
+ } else if (strcmp(key, "LogFile") == 0) {
+ if(!config->logfile) {
+ config->logfile = strdup(value);
+ pm_printf(PM_LOG_DEBUG, "config: logfile: %s\n", value);
+ }
+ } else if (strcmp(key, "XferCommand") == 0) {
+ config->xfercommand = strdup(value);
+ alpm_option_set_fetchcb(download_with_xfercommand);
+ pm_printf(PM_LOG_DEBUG, "config: xfercommand: %s\n", value);
+ } else if (strcmp(key, "CleanMethod") == 0) {
+ if (strcmp(value, "KeepInstalled") == 0) {
+ config->cleanmethod = PM_CLEAN_KEEPINST;
+ } else if (strcmp(value, "KeepCurrent") == 0) {
+ config->cleanmethod = PM_CLEAN_KEEPCUR;
+ } else {
+ pm_printf(PM_LOG_ERROR, _("invalid value for 'CleanMethod' : '%s'\n"), value);
+ return(1);
+ }
+ pm_printf(PM_LOG_DEBUG, "config: cleanmethod: %s\n", value);
+ } else {
+ pm_printf(PM_LOG_ERROR, _("directive '%s' with a value not recognized\n"), key);
+ return(1);
+ }
+
+ }
+ return(0);
+}
+
+static int _add_mirror(pmdb_t *db, char *value)
+{
+ const char *dbname = alpm_db_get_name(db);
+ /* let's attempt a replacement for the current repo */
+ char *temp = strreplace(value, "$repo", dbname);
+ /* let's attempt a replacement for the arch */
+ const char *arch = alpm_option_get_arch();
+ char *server;
+ if(arch) {
+ server = strreplace(temp, "$arch", arch);
+ free(temp);
+ } else {
+ if(strstr(temp, "$arch")) {
+ free(temp);
+ pm_printf(PM_LOG_ERROR, _("The mirror '%s' contains the $arch"
+ " variable, but no Architecture is defined.\n"), value);
+ return(1);
+ }
+ server = temp;
+ }
+
+ if(alpm_db_setserver(db, server) != 0) {
+ /* pm_errno is set by alpm_db_setserver */
+ pm_printf(PM_LOG_ERROR, _("could not add server URL to database '%s': %s (%s)\n"),
+ dbname, server, alpm_strerrorlast());
+ free(server);
+ return(1);
+ }
+
+ free(server);
+ return(0);
+}
+
/* The real parseconfig. Called with a null section argument by the publicly
* visible parseconfig so we can recall from within ourself on an include */
static int _parseconfig(const char *file, const char *givensection,
@@ -779,159 +900,70 @@ static int _parseconfig(const char *file, const char *givensection,
goto cleanup;
}
}
- } else {
- /* directive */
- char *key;
- /* strsep modifies the 'line' string: 'key \0 ptr' */
- key = line;
- ptr = line;
- strsep(&ptr, "=");
- strtrim(key);
- strtrim(ptr);
+ continue;
+ }

- if(key == NULL) {
- pm_printf(PM_LOG_ERROR, _("config file %s, line %d: syntax error in config file- missing key.\n"),
- file, linenum);
- ret = 1;
- goto cleanup;
- }
- /* For each directive, compare to the camelcase string. */
- if(section == NULL) {
- pm_printf(PM_LOG_ERROR, _("config file %s, line %d: All directives must belong to a section.\n"),
+ /* directive */
+ char *key, *value;
+ /* strsep modifies the 'line' string: 'key \0 value' */
+ key = line;
+ value = line;
+ strsep(&value, "=");
+ strtrim(key);
+ strtrim(value);
+
+ if(key == NULL) {
+ pm_printf(PM_LOG_ERROR, _("config file %s, line %d: syntax error in config file- missing key.\n"),
+ file, linenum);
+ ret = 1;
+ goto cleanup;
+ }
+ /* For each directive, compare to the camelcase string. */
+ if(section == NULL) {
+ pm_printf(PM_LOG_ERROR, _("config file %s, line %d: All directives must belong to a section.\n"),
+ file, linenum);
+ ret = 1;
+ goto cleanup;
+ }
+ if(strcmp(section, "options") == 0) {
+ if((ret = _parse_options(key, value)) != 0) {
+ pm_printf(PM_LOG_ERROR, _("config file %s, line %d: problem in options section\n"),
file, linenum);
ret = 1;
goto cleanup;
}
- if(ptr == NULL && strcmp(section, "options") == 0) {
- /* directives without settings, all in [options] */
- if(strcmp(key, "UseSyslog") == 0) {
- alpm_option_set_usesyslog(1);
- pm_printf(PM_LOG_DEBUG, "config: usesyslog\n");
- } else if(strcmp(key, "ILoveCandy") == 0) {
- config->chomp = 1;
- pm_printf(PM_LOG_DEBUG, "config: chomp\n");
- } else if(strcmp(key, "ShowSize") == 0) {
- config->showsize = 1;
- pm_printf(PM_LOG_DEBUG, "config: showsize\n");
- } else if(strcmp(key, "UseDelta") == 0) {
- alpm_option_set_usedelta(1);
- pm_printf(PM_LOG_DEBUG, "config: usedelta\n");
- } else if(strcmp(key, "TotalDownload") == 0) {
- config->totaldownload = 1;
- pm_printf(PM_LOG_DEBUG, "config: totaldownload\n");
- } else {
- pm_printf(PM_LOG_ERROR, _("config file %s, line %d: directive '%s' not recognized.\n"),
+ continue;
+ } else {
+ /* we are in a repo section */
+ if(strcmp(key, "Include") == 0) {
+ if(value == NULL) {
+ pm_printf(PM_LOG_ERROR, _("config file %s, line %d: directive %s needs a value\n"),
file, linenum, key);
ret = 1;
goto cleanup;
}
- } else {
- /* directives with settings */
- if(strcmp(key, "Include") == 0) {
- pm_printf(PM_LOG_DEBUG, "config: including %s\n", ptr);
- _parseconfig(ptr, section, db);
- /* Ignore include failures... assume non-critical */
- } else if(strcmp(section, "options") == 0) {
- if(strcmp(key, "NoUpgrade") == 0) {
- setrepeatingoption(ptr, "NoUpgrade", alpm_option_add_noupgrade);
- } else if(strcmp(key, "NoExtract") == 0) {
- setrepeatingoption(ptr, "NoExtract", alpm_option_add_noextract);
- } else if(strcmp(key, "IgnorePkg") == 0) {
- setrepeatingoption(ptr, "IgnorePkg", alpm_option_add_ignorepkg);
- } else if(strcmp(key, "IgnoreGroup") == 0) {
- setrepeatingoption(ptr, "IgnoreGroup", alpm_option_add_ignoregrp);
- } else if(strcmp(key, "HoldPkg") == 0) {
- setrepeatingoption(ptr, "HoldPkg", option_add_holdpkg);
- } else if(strcmp(key, "SyncFirst") == 0) {
- setrepeatingoption(ptr, "SyncFirst", option_add_syncfirst);
- } else if(strcmp(key, "Architecture") == 0) {
- if(!alpm_option_get_arch()) {
- setarch(ptr);
- }
- } else if(strcmp(key, "DBPath") == 0) {
- /* don't overwrite a path specified on the command line */
- if(!config->dbpath) {
- config->dbpath = strdup(ptr);
- pm_printf(PM_LOG_DEBUG, "config: dbpath: %s\n", ptr);
- }
- } else if(strcmp(key, "CacheDir") == 0) {
- if(alpm_option_add_cachedir(ptr) != 0) {
- pm_printf(PM_LOG_ERROR, _("problem adding cachedir '%s' (%s)\n"),
- ptr, alpm_strerrorlast());
- ret = 1;
- goto cleanup;
- }
- pm_printf(PM_LOG_DEBUG, "config: cachedir: %s\n", ptr);
- } else if(strcmp(key, "RootDir") == 0) {
- /* don't overwrite a path specified on the command line */
- if(!config->rootdir) {
- config->rootdir = strdup(ptr);
- pm_printf(PM_LOG_DEBUG, "config: rootdir: %s\n", ptr);
- }
- } else if (strcmp(key, "LogFile") == 0) {
- if(!config->logfile) {
- config->logfile = strdup(ptr);
- pm_printf(PM_LOG_DEBUG, "config: logfile: %s\n", ptr);
- }
- } else if (strcmp(key, "XferCommand") == 0) {
- config->xfercommand = strdup(ptr);
- alpm_option_set_fetchcb(download_with_xfercommand);
- pm_printf(PM_LOG_DEBUG, "config: xfercommand: %s\n", ptr);
- } else if (strcmp(key, "CleanMethod") == 0) {
- if (strcmp(ptr, "KeepInstalled") == 0) {
- config->cleanmethod = PM_CLEAN_KEEPINST;
- } else if (strcmp(ptr, "KeepCurrent") == 0) {
- config->cleanmethod = PM_CLEAN_KEEPCUR;
- } else {
- pm_printf(PM_LOG_ERROR, _("invalid value for 'CleanMethod' : '%s'\n"), ptr);
- ret = 1;
- goto cleanup;
- }
- pm_printf(PM_LOG_DEBUG, "config: cleanmethod: %s\n", ptr);
- } else {
- pm_printf(PM_LOG_ERROR, _("config file %s, line %d: directive '%s' not recognized.\n"),
- file, linenum, key);
- ret = 1;
- goto cleanup;
- }
- } else if(strcmp(key, "Server") == 0) {
- /* let's attempt a replacement for the current repo */
- char *temp = strreplace(ptr, "$repo", section);
- /* let's attempt a replacement for the arch */
- const char *arch = alpm_option_get_arch();
- char *server;
- if(arch) {
- server = strreplace(temp, "$arch", arch);
- free(temp);
- } else {
- if(strstr(temp, "$arch")) {
- free(temp);
- pm_printf(PM_LOG_ERROR, _("The mirror '%s' contains the $arch"
- " variable, but no Architecture is defined.\n"), ptr);
- ret = 1;
- goto cleanup;
- }
- server = temp;
- }
-
- if(alpm_db_setserver(db, server) != 0) {
- /* pm_errno is set by alpm_db_setserver */
- pm_printf(PM_LOG_ERROR, _("could not add server URL to database '%s': %s (%s)\n"),
- alpm_db_get_name(db), server, alpm_strerrorlast());
- free(server);
- ret = 1;
- goto cleanup;
- }
-
- free(server);
- } else {
- pm_printf(PM_LOG_ERROR, _("config file %s, line %d: directive '%s' not recognized.\n"),
+ pm_printf(PM_LOG_DEBUG, "config: including %s\n", value);
+ _parseconfig(value, section, db);
+ /* Ignore include failures... assume non-critical */
+ } else if(strcmp(key, "Server") == 0) {
+ if(value == NULL) {
+ pm_printf(PM_LOG_ERROR, _("config file %s, line %d: directive %s needs a value\n"),
file, linenum, key);
ret = 1;
goto cleanup;
}
+ if(_add_mirror(db, value) != 0) {
+ ret = 1;
+ goto cleanup;
+ }
+ } else {
+ pm_printf(PM_LOG_ERROR, _("config file %s, line %d: directive '%s' in repository section '%s' not recognized.\n"),
+ file, linenum, key, section);
+ ret = 1;
+ goto cleanup;
}
}
+
}

cleanup:
--
1.6.5.4

[PATCH 1/3] fix a few warnings reported by clang

2009-12-05T14:49:05Z

- remove unused variables
- some more sanity checks
- safer printf

Signed-off-by: Xavier Chantry <shiningxc@...>
---
lib/libalpm/add.c | 2 ++
lib/libalpm/be_files.c | 2 +-
lib/libalpm/conflict.c | 2 +-
lib/libalpm/package.c | 3 +--
src/pacman/callback.c | 4 +++-
src/pacman/pacman.c | 2 +-
src/pacman/util.c | 2 +-
7 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/lib/libalpm/add.c b/lib/libalpm/add.c
index ebcd6a5..dc565ac 100644
--- a/lib/libalpm/add.c
+++ b/lib/libalpm/add.c
@@ -483,6 +483,8 @@ static int commit_single_pkg(pmpkg_t *newpkg, int pkg_current, int pkg_count,

ALPM_LOG_FUNC;

+ ASSERT(trans != NULL, RET_ERR(PM_ERR_TRANS_NULL, -1));
+
snprintf(scriptlet, PATH_MAX, "%s%s-%s/install",
_alpm_db_path(db), alpm_pkg_get_name(newpkg),
alpm_pkg_get_version(newpkg));
diff --git a/lib/libalpm/be_files.c b/lib/libalpm/be_files.c
index 7fc20b8..f685019 100644
--- a/lib/libalpm/be_files.c
+++ b/lib/libalpm/be_files.c
@@ -140,8 +140,8 @@ static int dirlist_from_fs(const char *syncdbpath, alpm_list_t **dirlist)
entry[len+1] = '\0';
*dirlist = alpm_list_add(*dirlist, entry);
}
+ closedir(dbdir);
}
- closedir(dbdir);

*dirlist = alpm_list_msort(*dirlist, alpm_list_count(*dirlist), _alpm_str_cmp);
return(0);
diff --git a/lib/libalpm/conflict.c b/lib/libalpm/conflict.c
index e934c01..85db83b 100644
--- a/lib/libalpm/conflict.c
+++ b/lib/libalpm/conflict.c
@@ -408,7 +408,7 @@ alpm_list_t *_alpm_db_find_fileconflicts(pmdb_t *db, pmtrans_t *trans,

ALPM_LOG_FUNC;

- if(db == NULL || upgrade == NULL) {
+ if(db == NULL || upgrade == NULL || trans == NULL) {
return(NULL);
}

diff --git a/lib/libalpm/package.c b/lib/libalpm/package.c
index 83a2fb8..10220af 100644
--- a/lib/libalpm/package.c
+++ b/lib/libalpm/package.c
@@ -459,7 +459,6 @@ void SYMEXPORT *alpm_pkg_changelog_open(pmpkg_t *pkg)
struct archive *archive = NULL;
struct archive_entry *entry;
const char *pkgfile = pkg->origin_data.file;
- int ret = ARCHIVE_OK;

if((archive = archive_read_new()) == NULL) {
RET_ERR(PM_ERR_LIBARCHIVE, NULL);
@@ -473,7 +472,7 @@ void SYMEXPORT *alpm_pkg_changelog_open(pmpkg_t *pkg)
RET_ERR(PM_ERR_PKG_OPEN, NULL);
}

- while((ret = archive_read_next_header(archive, &entry)) == ARCHIVE_OK) {
+ while(archive_read_next_header(archive, &entry) == ARCHIVE_OK) {
const char *entry_name = archive_entry_pathname(entry);

if(strcmp(entry_name, ".CHANGELOG") == 0) {
diff --git a/src/pacman/callback.c b/src/pacman/callback.c
index 858bfdf..15f5423 100644
--- a/src/pacman/callback.c
+++ b/src/pacman/callback.c
@@ -365,6 +365,8 @@ void cb_trans_progress(pmtransprog_t event, const char *pkgname, int percent,
case PM_TRANS_PROGRESS_CONFLICTS_START:
opr = _("checking for file conflicts");
break;
+ default:
+ return;
}

/* find # of digits in package counts to scale output */
@@ -504,7 +506,7 @@ void cb_dl_progress(const char *filename, off_t file_xfered, off_t file_total)
gettimeofday(&initial_time, NULL);
xfered_last = (off_t)0;
rate_last = 0.0;
- timediff = get_update_timediff(1);
+ get_update_timediff(1);
}
} else if(file_xfered == file_total) {
/* compute final values */
diff --git a/src/pacman/pacman.c b/src/pacman/pacman.c
index ff6ef5c..3dccd48 100644
--- a/src/pacman/pacman.c
+++ b/src/pacman/pacman.c
@@ -217,7 +217,7 @@ static void setarch(const char *arch)
static void cleanup(int ret) {
/* free alpm library resources */
if(alpm_release() == -1) {
- pm_printf(PM_LOG_ERROR, alpm_strerrorlast());
+ pm_printf(PM_LOG_ERROR, "%s\n", alpm_strerrorlast());
}

/* free memory */
diff --git a/src/pacman/util.c b/src/pacman/util.c
index 115b367..d395f3d 100644
--- a/src/pacman/util.c
+++ b/src/pacman/util.c
@@ -219,7 +219,7 @@ void indentprint(const char *str, int indent)
p = wcstr;
cidx = indent;

- if(!p) {
+ if(!p || !len) {
return;
}

--
1.6.5.4

Re: [PATCH] makepkg: document environment variables PKGDEST and SRCDEST

2009-12-04T20:45:12Z

Cedric Staniewski wrote:
> Signed-off-by: Cedric Staniewski <cedric@...>
> ---

Good. Pushed to my working branch.

Allan

[PATCH] makepkg: document environment variables PKGDEST and SRCDEST

2009-12-04T08:49:54Z

Signed-off-by: Cedric Staniewski <cedric@...>
---
doc/makepkg.8.txt | 8 ++++++++
1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
index 72206d8..63a2cf4 100644
--- a/doc/makepkg.8.txt
+++ b/doc/makepkg.8.txt
@@ -170,6 +170,14 @@ Environment Variables
must be supported by this command. If the variable is not set or
empty, makepkg will fall back to `pacman'.

+**PKGDEST=**"/path/to/folder"::
+ Folder where the resulting packages will be stored. Overrides the
+ corresponding value defined in linkman:makepkg.conf[5].
+
+**SRCDEST=**"/path/to/folder"::
+ Folder where the downloaded sources will be stored. Overrides the
+ corresponding value defined in linkman:makepkg.conf[5].
+

Additional Features
-------------------
--
1.6.5.4

[PATCH] makepkg: improve removal of installed dependencies

2009-12-03T06:33:24Z

Compare a list of packages on the system before and after dependency resolution
in order to get a complete list of packages to remove. This allows makepkg to
remove packages installed due to provides.

Bail in cases where packages that were on the system originally have been
removed as there is a risk of breaking the system when removing the new
packages

Fixes FS#15144

Signed-off-by: Allan McRae <allan@...>
---

This moves the generation of the installed package list to directly after
the installation of deps to prevent issues if the user installs packages
while the build is occurring:
http://bugs.archlinux.org/task/15144#comment53938

Rebased on Cedric's run_pacman and $PACMAN patches.

doc/makepkg.8.txt | 6 +++---
scripts/makepkg.sh.in | 39 +++++++++++++++++++++------------------
2 files changed, 24 insertions(+), 21 deletions(-)

diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
index ad27bca..168b369 100644
--- a/doc/makepkg.8.txt
+++ b/doc/makepkg.8.txt
@@ -166,9 +166,9 @@ Environment Variables
---------------------
*PACMAN*::
The command that will be used to check for missing dependencies and to
- install and remove packages. Pacman's -U, -T, -S and -Rns operations
- must be supported by this command. If the variable is not set or
- empty, makepkg will fall back to `pacman'.
+ install and remove packages. Pacman's -Qq, -Rns, -S, -T, and -U
+ operations must be supported by this command. If the variable is not
+ set or empty, makepkg will fall back to `pacman'.

Additional Features
diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in
index c730150..5743ea4 100644
--- a/scripts/makepkg.sh.in
+++ b/scripts/makepkg.sh.in
@@ -27,8 +27,10 @@

# makepkg uses quite a few external programs during its execution. You
# need to have at least the following installed for makepkg to function:
-# bsdtar (libarchive), bzip2, coreutils, fakeroot, find (findutils),
-# getopt (util-linux), gettext, grep, gzip, openssl, sed, tput (ncurses)
+# bsdtar (libarchive), bzip2, coreutils, diff (diffutils), fakeroot,
+# find (findutils), getopt (util-linux), gettext, grep, gzip, openssl,
+# sed, tput (ncurses)
+

# gettext initialization
export TEXTDOMAIN='pacman'
@@ -343,7 +345,7 @@ download_file() {

run_pacman() {
local ret=0
- if (( ! ASROOT )) && [[ $1 != "-T" ]] && sudo -l $PACMAN &>/dev/null; then
+ if (( ! ASROOT )) && [[ $1 != "-T" && $1 != "-Qq" ]] && sudo -l $PACMAN &>/dev/null; then
sudo $PACMAN $PACMAN_OPTS "$@" || ret=$?
else
$PACMAN $PACMAN_OPTS "$@" || ret=$?
@@ -398,7 +400,6 @@ handle_deps() {
}

resolve_deps() {
- # $pkgdeps is a GLOBAL variable, used by remove_deps()
local R_DEPS_SATISFIED=0
local R_DEPS_MISSING=1

@@ -408,7 +409,6 @@ resolve_deps() {
fi

if handle_deps $deplist; then
- pkgdeps="$pkgdeps $deplist"
# check deps again to make sure they were resolved
deplist="$(check_deps $*)"
[[ -z $deplist ]] && return $R_DEPS_SATISFIED
@@ -425,23 +425,25 @@ resolve_deps() {
return $R_DEPS_MISSING
}

-# fix flyspray bug #5923
remove_deps() {
- # $pkgdeps is a GLOBAL variable, set by resolve_deps()
(( ! RMDEPS )) && return
- [[ -z $pkgdeps ]] && return

- local dep depstrip deplist
- deplist=""
- for dep in $pkgdeps; do
- depstrip="${dep%%[<=>]*}"
- deplist="$deplist $depstrip"
- done
+ # check for packages removed during dependency install (e.g. due to conflicts)
+ # removing all installed packages is risky in this case
+ if (( $(diff <(printf "%s\n" "${original_pkglist[@]}") \
+ <(printf "%s\n" "${current_pkglist[@]}") | grep "<" | wc -l) > 0 )); then
+ warning "$(gettext "Failed to remove installed dependencies.")"
+ return 0
+ fi

- msg "Removing installed dependencies..."
+ local deplist=($(diff <(printf "%s\n" "${original_pkglist[@]}") \
+ <(printf "%s\n" "${current_pkglist[@]}") | grep ">"))
+ deplist=(${deplist[@]#>})
+ [ ${#deplist[@]} -eq 0 ] && return

+ msg "Removing installed dependencies..."
# exit cleanly on failure to remove deps as package has been built successfully
- if ! run_pacman -Rns $deplist; then
+ if ! run_pacman -Rn ${deplist[@]}; then
warning "$(gettext "Failed to remove installed dependencies.")"
return 0
fi
@@ -1874,14 +1876,13 @@ if (( SOURCEONLY )); then
exit 0
fi

-# fix flyspray bug #5973
if (( NODEPS || NOBUILD || REPKG )); then
# no warning message needed for nobuild, repkg
if (( NODEPS )); then
warning "$(gettext "Skipping dependency checks.")"
fi
elif [ $(type -p "${PACMAN%% *}") ]; then
- unset pkgdeps # Set by resolve_deps() and used by remove_deps()
+ original_pkglist=($(run_pacman -Qq)) # required by remove_deps
deperr=0

msg "$(gettext "Checking Runtime Dependencies...")"
@@ -1890,6 +1891,8 @@ elif [ $(type -p "${PACMAN%% *}") ]; then
msg "$(gettext "Checking Buildtime Dependencies...")"
resolve_deps ${makedepends[@]} || deperr=1

+ current_pkglist=($(run_pacman -Qq)) # required by remove_deps
+
if (( deperr )); then
error "$(gettext "Could not resolve all dependencies.")"
exit 1
--
1.6.5.3

Re: [PATCH 1/3] makepkg: allow to specify an alternative pacman command

2009-12-03T05:55:18Z

Dan McGee wrote:

> On Thu, Dec 3, 2009 at 6:09 AM, Allan McRae <allan@...> wrote:
>> Cedric Staniewski wrote:
>>> If PACMAN environment variable is set, makepkg will try to use this
>>> command to check for installed dependencies and to install or remove
>>> packages. Otherwise, makepkg will fall back to pacman.
>>>
>>> Implements FS#13028.
>>>
>>> Signed-off-by: Cedric Staniewski <cedric@...>
>>> ---
>>> doc/makepkg.8.txt | 9 +++++++++
>>> scripts/makepkg.sh.in | 19 +++++++++++--------
>>> 2 files changed, 20 insertions(+), 8 deletions(-)
>>>
>>> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
>>> index 18ee6ed..703c1b0 100644
>>> --- a/doc/makepkg.8.txt
>>> +++ b/doc/makepkg.8.txt
>>> @@ -162,6 +162,15 @@ Options
>>> useful if you are redirecting makepkg output to file.
>>> +Environment Variables
>> Useful Environment Variables? The lead in sounds better, but maybe not
>> quite right. Not sure here...
>
> "Environment Variables" is a standard manpage header, usually found
> near the bottom. I didn't look where this patch actually put it in the
> manpage, but see examples like less, grep, etc.

OK, fine. It is place after OPTIONS and before ADDITIONAL FEATURES and
CONFIGURATION. Seems a reasonable place.

>>> +---------------------
>>> +*PACMAN*::
>>> + The command which will be used to check for missing dependencies
>>> and to
>>> + install and remove packages. Pacman's -U, -T, -S and -Rns
>>> operations
>>> + must be supported by this command. If the variable is not set or
>>> + empty, makepkg will fall back to `pacman'.
>> The command _that_ will... The rest of the description sounds fine.
I will change the "which" to "that" and pull to my working branch.

Allan

Re: [PATCH 1/3] makepkg: allow to specify an alternative pacman command

2009-12-03T05:36:06Z

On Thu, Dec 3, 2009 at 6:09 AM, Allan McRae <allan@...> wrote:

> Cedric Staniewski wrote:
>>
>> If PACMAN environment variable is set, makepkg will try to use this
>> command to check for installed dependencies and to install or remove
>> packages. Otherwise, makepkg will fall back to pacman.
>>
>> Implements FS#13028.
>>
>> Signed-off-by: Cedric Staniewski <cedric@...>
>> ---
>> doc/makepkg.8.txt | 9 +++++++++
>> scripts/makepkg.sh.in | 19 +++++++++++--------
>> 2 files changed, 20 insertions(+), 8 deletions(-)
>>
>> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
>> index 18ee6ed..703c1b0 100644
>> --- a/doc/makepkg.8.txt
>> +++ b/doc/makepkg.8.txt
>> @@ -162,6 +162,15 @@ Options
>> useful if you are redirecting makepkg output to file.
>> +Environment Variables
>
> Useful Environment Variables? The lead in sounds better, but maybe not
> quite right. Not sure here...

"Environment Variables" is a standard manpage header, usually found
near the bottom. I didn't look where this patch actually put it in the
manpage, but see examples like less, grep, etc.

>
>> +---------------------
>> +*PACMAN*::
>> + The command which will be used to check for missing dependencies
>> and to
>> + install and remove packages. Pacman's -U, -T, -S and -Rns
>> operations
>> + must be supported by this command. If the variable is not set or
>> + empty, makepkg will fall back to `pacman'.
>
> The command _that_ will... The rest of the description sounds fine.
>
> Otherwise, the patch is good.
>
>
>

Re: [PATCH 1/3] makepkg: allow to specify an alternative pacman command

2009-12-03T04:09:09Z

Cedric Staniewski wrote:

> If PACMAN environment variable is set, makepkg will try to use this
> command to check for installed dependencies and to install or remove
> packages. Otherwise, makepkg will fall back to pacman.
>
> Implements FS#13028.
>
> Signed-off-by: Cedric Staniewski <cedric@...>
> ---
> doc/makepkg.8.txt | 9 +++++++++
> scripts/makepkg.sh.in | 19 +++++++++++--------
> 2 files changed, 20 insertions(+), 8 deletions(-)
>
> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
> index 18ee6ed..703c1b0 100644
> --- a/doc/makepkg.8.txt
> +++ b/doc/makepkg.8.txt
> @@ -162,6 +162,15 @@ Options
> useful if you are redirecting makepkg output to file.
>
>
> +Environment Variables

Useful Environment Variables? The lead in sounds better, but maybe not
quite right. Not sure here...

> +---------------------
> +*PACMAN*::
> + The command which will be used to check for missing dependencies and to
> + install and remove packages. Pacman's -U, -T, -S and -Rns operations
> + must be supported by this command. If the variable is not set or
> + empty, makepkg will fall back to `pacman'.

The command _that_ will... The rest of the description sounds fine.

Otherwise, the patch is good.

Re: [PATCH] makepkg: remove srclinks directory on error exit

2009-12-03T01:43:38Z

Cedric Staniewski wrote:

> On 12/03/2009 05:08 AM, Allan McRae wrote:
>> Cedric Staniewski wrote:
>>> When makepkg exits in create_srcpackage(), the (temporary) srclinks
>>> directory is left behind.
>>>
>>> Signed-off-by: Cedric Staniewski <cedric@...>
>>> ---
>>> scripts/makepkg.sh.in | 1 +
>>> 1 files changed, 1 insertions(+), 0 deletions(-)
>>>
>>> diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in
>>> index ceaa8a6..b618881 100644
>>> --- a/scripts/makepkg.sh.in
>>> +++ b/scripts/makepkg.sh.in
>>> @@ -118,6 +118,7 @@ trap_exit() {
>>> echo
>>> error "$@"
>>> fi
>>> + [[ -n $srclinks ]] && rm -rf "$srclinks"
>>> exit 1
>>> }
>>>
>> I am not sure about this as if the source package creation fails, I
>> would like to be able to look in the directory to track down why. Much
>> like how "-c" does not clear the source directory on a build failure. So
>> my initial reaction is a -1.
>>
>> As an aside, did you actually encounter an error during source package
>> creation or just note that could be left behind from the code?
>>
>
> I actually did not encounter an error but aborted via ctrl+c. Keeping
> this folder is useless in my opinion as there are error messages for
> pretty much everything and the three different external commands which
> are used (mkdir, ln, bsdtar) usually throw meaningful error messages,
> too. Aside from that this folder only contains smylinks anyway and I
> doubt they are useful for tracking down errors.
>
> Another alternative would be introducing a new cleanup function which is
> only run on SIGINT and moving this line there, though.
>

OK, your comment about the external commands being used having good
error messages has me convinced something needs done. It just seems a
strange place for this particular clean up. Saying that, I'm not sure
what would be a better place unless we had an error trap just for that
function which passed on to trap_exit.

Dan, Xavier, others: any comments?

Re: [PATCH 3/3] makepkg: document environment variables PKGDEST and SRCDEST

2009-12-03T01:17:55Z

Cedric Staniewski wrote:

> On 12/03/2009 05:19 AM, Allan McRae wrote:
>> Cedric Staniewski wrote:
>>> Signed-off-by: Cedric Staniewski <cedric@...>
>>> ---
>>> doc/makepkg.8.txt | 15 +++++++++++++++
>>> 1 files changed, 15 insertions(+), 0 deletions(-)
>>>
>>> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
>>> index 703c1b0..ccb9a28 100644
>>> --- a/doc/makepkg.8.txt
>>> +++ b/doc/makepkg.8.txt
>>> @@ -170,6 +170,21 @@ Environment Variables
>>> must be supported by this command. If the variable is not set or
>>> empty, makepkg will fall back to `pacman'.
>>>
>>> +**PKGDEST=**"/path/to/folder"::
>>> + If this value is not set, packages will by default be placed in the
>>> + current directory (location of the linkman:PKGBUILD[5]). Many people
>>> + like to keep all their packages in one place so this option allows
>>> + this behavior. A common location is ``/home/packages''.
>>> + This environment variable will override the corresponding value
>>> + defined in linkman:makepkg.conf[5].
>> I do not like the repetition of makepkg.conf.5 here. I'd prefer just
>> **PKGDEST=**"/path/to/folder"::
>> Overrides the corresponding value defined in linkman:makepkg.conf[5].
>>
>> Or something quite simple like that. People can then look up
>> makepkg.conf.5 to find out what the variable does.
>>
>>
>
> I agree with you. How about
>
> **PKGDEST=**"/path/to/folder"::
> Folder where the resulting packages will be stored. Overrides the
> corresponding value defined in linkman:makepkg.conf[5].
>
> **SRCDEST=**"/path/to/folder"::
> Folder where the downloaded sources will be stored. Overrides the
> corresponding value defined in linkman:makepkg.conf[5].
>

Looks good to me. I'll let Dan comment as he always picks holes in my
documentation! :P

Re: [PATCH] makepkg: remove srclinks directory on error exit

2009-12-03T00:30:52Z

On 12/03/2009 05:08 AM, Allan McRae wrote:

> Cedric Staniewski wrote:
>> When makepkg exits in create_srcpackage(), the (temporary) srclinks
>> directory is left behind.
>>
>> Signed-off-by: Cedric Staniewski <cedric@...>
>> ---
>> scripts/makepkg.sh.in | 1 +
>> 1 files changed, 1 insertions(+), 0 deletions(-)
>>
>> diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in
>> index ceaa8a6..b618881 100644
>> --- a/scripts/makepkg.sh.in
>> +++ b/scripts/makepkg.sh.in
>> @@ -118,6 +118,7 @@ trap_exit() {
>> echo
>> error "$@"
>> fi
>> + [[ -n $srclinks ]] && rm -rf "$srclinks"
>> exit 1
>> }
>>
>
> I am not sure about this as if the source package creation fails, I
> would like to be able to look in the directory to track down why. Much
> like how "-c" does not clear the source directory on a build failure. So
> my initial reaction is a -1.
>
> As an aside, did you actually encounter an error during source package
> creation or just note that could be left behind from the code?
>

I actually did not encounter an error but aborted via ctrl+c. Keeping
this folder is useless in my opinion as there are error messages for
pretty much everything and the three different external commands which
are used (mkdir, ln, bsdtar) usually throw meaningful error messages,
too. Aside from that this folder only contains smylinks anyway and I
doubt they are useful for tracking down errors.

Another alternative would be introducing a new cleanup function which is
only run on SIGINT and moving this line there, though.

Re: [PATCH 3/3] makepkg: document environment variables PKGDEST and SRCDEST

2009-12-03T00:09:29Z

On 12/03/2009 05:19 AM, Allan McRae wrote:

> Cedric Staniewski wrote:
>> Signed-off-by: Cedric Staniewski <cedric@...>
>> ---
>> doc/makepkg.8.txt | 15 +++++++++++++++
>> 1 files changed, 15 insertions(+), 0 deletions(-)
>>
>> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
>> index 703c1b0..ccb9a28 100644
>> --- a/doc/makepkg.8.txt
>> +++ b/doc/makepkg.8.txt
>> @@ -170,6 +170,21 @@ Environment Variables
>> must be supported by this command. If the variable is not set or
>> empty, makepkg will fall back to `pacman'.
>>
>> +**PKGDEST=**"/path/to/folder"::
>> + If this value is not set, packages will by default be placed in the
>> + current directory (location of the linkman:PKGBUILD[5]). Many people
>> + like to keep all their packages in one place so this option allows
>> + this behavior. A common location is ``/home/packages''.
>> + This environment variable will override the corresponding value
>> + defined in linkman:makepkg.conf[5].
>
> I do not like the repetition of makepkg.conf.5 here. I'd prefer just
> **PKGDEST=**"/path/to/folder"::
> Overrides the corresponding value defined in linkman:makepkg.conf[5].
>
> Or something quite simple like that. People can then look up
> makepkg.conf.5 to find out what the variable does.
>
>

I agree with you. How about

**PKGDEST=**"/path/to/folder"::
Folder where the resulting packages will be stored. Overrides the
corresponding value defined in linkman:makepkg.conf[5].

**SRCDEST=**"/path/to/folder"::
Folder where the downloaded sources will be stored. Overrides the
corresponding value defined in linkman:makepkg.conf[5].

>> +
>> +**SRCDEST=**"/path/to/folder"::
>> + If this value is not set, downloaded source files will only be
>> stored
>> + in the current directory. Many people like to keep all source
>> files in
>> + a central location for easy cleanup, so this path can be set here.
>> + This environment variable will override the corresponding value
>> + defined in linkman:makepkg.conf[5].
>> +
>>
>> Additional Features
>> -------------------
>

Re: [PATCH] docs: set text width to 72 in vim directive

2009-12-02T23:59:02Z

On 12/03/2009 01:47 AM, Dan McGee wrote:
> On Wed, Dec 2, 2009 at 12:05 PM, Cedric Staniewski <cedric@...> wrote:
>
> A "why" in the commit message would be awfully nice, I'm confused here. :)
>
>> Signed-off-by: Cedric Staniewski <cedric@...>

Sorry, I thought it were quite clear. The text in the man pages is
already wrapped after a certain amount of characters and I thought is
would be nice to add this to the vim directive or am I missing something
here?

Re: inside PKGBUILD customizable makepkg PURGE_TARGETS

2009-12-02T21:10:43Z

Ciprian Dorin, Craciun wrote:

> On Thu, Dec 3, 2009 at 2:39 AM, Dan McGee <dpmcgee@...> wrote:
>> On Wed, Dec 2, 2009 at 3:48 PM, Allan McRae <allan@...> wrote:
>>> Ciprian Dorin, Craciun wrote:
>>>> Hello all!
>>>>
>>>> Again I have a proposal to enhance makepkg PURGE_TARGETS handling.
>>>>
>>>> I think that enabling customizing PURGE_TARGETS inside PKGBUILD
>>>> would be a benefit in the following context:
>>> Without checking, I would have thought that exporting "PURGE_TRAGETS=..."
>>> within a package function would just work. I guess the old value would not
>>> be restored afterwards though.
>> This seems like complete overkill for makepkg to support. I'm -1 on this.
>
>
> What overkill? Really? It's a 3 or 4-line update:
>
> ~~~~ instead of this:
> if [ "$(check_option purge)" = "y" -a -n "${PURGE_TARGETS[*]}" ]; then
> msg2 "$(gettext "Purging other files...")"
> ~~~~ we'll have this:
> + local purge_targets=( "${pkg_purge_targets[@]:-${PURGE_TARGETS[@]}}" )
> + eval 'purge_targets=(
> "${'"${pkgname}"'_purge_targets[@]:-${purge_targets[@]}" )'
> + if [ "$(check_option purge)" = "y" -a -n "${purge_targets[*]}" ]; then
> - if [ "$(check_option purge)" = "y" -a -n "${PURGE_TARGETS[*]}" ]; then
> msg2 "$(gettext "Purging other files...")"
> ~~~~ and we have to replace `PURGE_TARGETS` with `purge_targets`
> somewhare below in 2 places.
>

As I said, defining PURGE_TARGETS in a package_foo() function already works:

e.g.

package_t1() {
PURGE_TARGETS=(*.a)
touch $pkgdir/foo.a
touch $pkgdir/foo.doc
}

package_t2() {
PURGE_TARGETS=(*.doc)
touch $pkgdir/foo.a
touch $pkgdir/foo.doc
}

Guess which file ends up in which package...

>
> Guys, as developers of makepkg, you really have something against
> extensions that would allow writing simpler or more versatile
> PKGBUILD's...

No, we just do not believe any of the ideas you have shown us so far are
making simpler PKGBUILDs or achieving something that can not already be
done with makepkg using the same number of lines in the PKGBUILD

> For example, below is one of my PKGBUILDS, that uses what I call a
> "template", which allows me to abstract away patching, calling
> configure, and make. And in order to support this split of packages,
> all I have to do is update my `configure+make+make-install` script to
> understand this, and just put `pkgname=( isync-bin isync-doc )`.
> (Unfortunately this doesn't have a lib or dev).
>
> My `configure+make+make-install` just implements `build` and
> `package` functions that use the variables defined in the PKGBUILD,
> with some sensible defaults. (I think this resembles somehow GoboLinux
> recipies...)
>
> ~~~~
> . "${pkgbuild_tools_root}/configure+make+make-install"
>
> pkgname=( isync )
> _pkgver=1.0.4
> _pkgpatch=+1
> pkgver="${_pkgver}${_pkgpatch}"
> pkgrel=1
>
> depends=()
> makedepends=(
> openssl db
> gcc binutils glibc
> make coreutils bash
> sed grep patch
> )
>
> _declare_arch
> _declare_paths
> _declare_build_environment
> _declare_configure_options
>
> _configure_options=(
> "${_configure_options[@]}"
> --disable-compat
> )
>
> _LIBS=(
> "${_LIBS[@]}"
> -static
> -ldb -lpthread
> )
>
> _src_patches=(
> imap_recursive_maildir.patch
> imap_certificate_ignore.patch
> )
>
> source=(
> "http://switch.dl.sourceforge.net/project/isync/isync/${_pkgver}/isync-${_pkgver}.tar.gz"
> imap_recursive_maildir.patch
> imap_certificate_ignore.patch
> )
> md5sums=(
> 8a836a6f4b43cd38a8b8153048417616
> fd42b61f2979d2bf2d08f1528a292ec5
> efbe84127425024676b07c2fdef87a98
> )

I would not call that simpler. I look at that and have only vague ideas
what the packaging would be doing, whereas a normal package I see the
exact commands.

Allan

Re: inside PKGBUILD customizable makepkg PURGE_TARGETS

2009-12-02T20:56:05Z

On Thu, Dec 3, 2009 at 2:39 AM, Dan McGee <dpmcgee@...> wrote:

> On Wed, Dec 2, 2009 at 3:48 PM, Allan McRae <allan@...> wrote:
>> Ciprian Dorin, Craciun wrote:
>>>
>>> Hello all!
>>>
>>> Again I have a proposal to enhance makepkg PURGE_TARGETS handling.
>>>
>>> I think that enabling customizing PURGE_TARGETS inside PKGBUILD
>>> would be a benefit in the following context:
>>
>> Without checking, I would have thought that exporting "PURGE_TRAGETS=..."
>> within a package function would just work. I guess the old value would not
>> be restored afterwards though.
>
> This seems like complete overkill for makepkg to support. I'm -1 on this.

What overkill? Really? It's a 3 or 4-line update:

~~~~ instead of this:
if [ "$(check_option purge)" = "y" -a -n "${PURGE_TARGETS[*]}" ]; then
msg2 "$(gettext "Purging other files...")"
~~~~ we'll have this:
+ local purge_targets=( "${pkg_purge_targets[@]:-${PURGE_TARGETS[@]}}" )
+ eval 'purge_targets=(
"${'"${pkgname}"'_purge_targets[@]:-${purge_targets[@]}" )'
+ if [ "$(check_option purge)" = "y" -a -n "${purge_targets[*]}" ]; then
- if [ "$(check_option purge)" = "y" -a -n "${PURGE_TARGETS[*]}" ]; then
msg2 "$(gettext "Purging other files...")"
~~~~ and we have to replace `PURGE_TARGETS` with `purge_targets`
somewhare below in 2 places.

> If upstream builds can't get their act together and provide individual
> make commands, file a issue upstream. If that doesn't work, then
> PURGE_TARGETS is really no more than a

Have I mentioned that even pacman doesn't support these make commands?

> for file in glob_exp; do
> rm -f file
> done
>
> which is really just a one-liner in a PKGBUILD anyway. Why complicate
> the matter?

Actually it's a 3-liner per `something_package` function, in every
PKGBUILD, instead of a true 1-liner in every function in every
PKGBUILD.

> -Dan

Guys, as developers of makepkg, you really have something against
extensions that would allow writing simpler or more versatile
PKGBUILD's...

For example, below is one of my PKGBUILDS, that uses what I call a
"template", which allows me to abstract away patching, calling
configure, and make. And in order to support this split of packages,
all I have to do is update my `configure+make+make-install` script to
understand this, and just put `pkgname=( isync-bin isync-doc )`.
(Unfortunately this doesn't have a lib or dev).

My `configure+make+make-install` just implements `build` and
`package` functions that use the variables defined in the PKGBUILD,
with some sensible defaults. (I think this resembles somehow GoboLinux
recipies...)

~~~~
. "${pkgbuild_tools_root}/configure+make+make-install"

pkgname=( isync )
_pkgver=1.0.4
_pkgpatch=+1
pkgver="${_pkgver}${_pkgpatch}"
pkgrel=1

depends=()
makedepends=(
openssl db
gcc binutils glibc
make coreutils bash
sed grep patch
)

_declare_arch
_declare_paths
_declare_build_environment
_declare_configure_options

_configure_options=(
"${_configure_options[@]}"
--disable-compat
)

_LIBS=(
"${_LIBS[@]}"
-static
-ldb -lpthread
)

_src_patches=(
imap_recursive_maildir.patch
imap_certificate_ignore.patch
)

source=(
"http://switch.dl.sourceforge.net/project/isync/isync/${_pkgver}/isync-${_pkgver}.tar.gz"
imap_recursive_maildir.patch
imap_certificate_ignore.patch
)
md5sums=(
8a836a6f4b43cd38a8b8153048417616
fd42b61f2979d2bf2d08f1528a292ec5
efbe84127425024676b07c2fdef87a98
)
~~~~

But as I said, no problem, I'll just hack my own makepkg that
understands something similar with PKGBUILDS, and extended with these
recipies.

Thanks,
Ciprian.

Re: [PATCH 3/3] makepkg: document environment variables PKGDEST and SRCDEST

2009-12-02T20:19:34Z

Cedric Staniewski wrote:

> Signed-off-by: Cedric Staniewski <cedric@...>
> ---
> doc/makepkg.8.txt | 15 +++++++++++++++
> 1 files changed, 15 insertions(+), 0 deletions(-)
>
> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
> index 703c1b0..ccb9a28 100644
> --- a/doc/makepkg.8.txt
> +++ b/doc/makepkg.8.txt
> @@ -170,6 +170,21 @@ Environment Variables
> must be supported by this command. If the variable is not set or
> empty, makepkg will fall back to `pacman'.
>
> +**PKGDEST=**"/path/to/folder"::
> + If this value is not set, packages will by default be placed in the
> + current directory (location of the linkman:PKGBUILD[5]). Many people
> + like to keep all their packages in one place so this option allows
> + this behavior. A common location is ``/home/packages''.
> + This environment variable will override the corresponding value
> + defined in linkman:makepkg.conf[5].

I do not like the repetition of makepkg.conf.5 here. I'd prefer just
**PKGDEST=**"/path/to/folder"::
Overrides the corresponding value defined in linkman:makepkg.conf[5].

Or something quite simple like that. People can then look up
makepkg.conf.5 to find out what the variable does.

> +
> +**SRCDEST=**"/path/to/folder"::
> + If this value is not set, downloaded source files will only be stored
> + in the current directory. Many people like to keep all source files in
> + a central location for easy cleanup, so this path can be set here.
> + This environment variable will override the corresponding value
> + defined in linkman:makepkg.conf[5].
> +
>
> Additional Features
> -------------------

Re: [PATCH] makepkg: remove srclinks directory on error exit

2009-12-02T20:08:50Z

Cedric Staniewski wrote:

> When makepkg exits in create_srcpackage(), the (temporary) srclinks
> directory is left behind.
>
> Signed-off-by: Cedric Staniewski <cedric@...>
> ---
> scripts/makepkg.sh.in | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in
> index ceaa8a6..b618881 100644
> --- a/scripts/makepkg.sh.in
> +++ b/scripts/makepkg.sh.in
> @@ -118,6 +118,7 @@ trap_exit() {
> echo
> error "$@"
> fi
> + [[ -n $srclinks ]] && rm -rf "$srclinks"
> exit 1
> }
>

I am not sure about this as if the source package creation fails, I
would like to be able to look in the directory to track down why. Much
like how "-c" does not clear the source directory on a build failure.
So my initial reaction is a -1.

As an aside, did you actually encounter an error during source package
creation or just note that could be left behind from the code?

Allan

Re: [PATCH] makepkg: check for references to build root in package

2009-12-02T17:24:57Z

Dan McGee wrote:

> On Wed, Dec 2, 2009 at 12:17 PM, Cedric Staniewski <cedric@...> wrote:
>> On 10/28/2009 10:07 PM, Jeff wrote:
>>> On Mon, Oct 26, 2009 at 09:08:21AM +1000, Allan McRae wrote:
>>>> I have adjusted the patch to use Cedric's syntax now that it is working
>>>> for me.
>>> FWIW, you can also make the line a tad shorter by using grep's -q
>>> instead of &>/dev/null.
>>>
>>
>> I noticed this patch (latest revision[1]) has not made it into master yet...
>>
>> [1]
>> http://projects.archlinux.org/users/allan/pacman.git/commit/?id=ce40da51611bf6dc236d814894fc90bd624bdc2f
>
> I'm not convinced on this one as it has to scan the entire pkg/
> directory, which for some things, could be gigantic (icc,
> go-openoffice, etc.). Am I being over-performance crazy? I guess the
> zip is going to do it too...
>
> To me this seems way outside the realm of makepkg and belongs in namcap.
>

The problem with this is that namcap does not know anything about
$srcdir (and how can it) so cannot check for references to it. This is
unlike the check for missing backup files that is already in
check_package() which should probably be in namcap...

I also tested this in large packages (e.g. openoffice) and on my system
(2GHz) it takes ~10sec which is a very small fraction of the build time.

As an aside, extracting and grepping through the entire core+extra+some
of community takes me ~10min. Per package this is a tiny amount of time
and includes extraction of tarball and not having files in the cache.

Allan

Re: [PATCH] makepkg: check for references to build root in package

2009-12-02T16:50:17Z

On Wed, Dec 2, 2009 at 12:17 PM, Cedric Staniewski <cedric@...> wrote:

> On 10/28/2009 10:07 PM, Jeff wrote:
>> On Mon, Oct 26, 2009 at 09:08:21AM +1000, Allan McRae wrote:
>>>
>>> I have adjusted the patch to use Cedric's syntax now that it is working
>>> for me.
>>
>> FWIW, you can also make the line a tad shorter by using grep's -q
>> instead of &>/dev/null.
>>
>
>
> I noticed this patch (latest revision[1]) has not made it into master yet...
>
> [1]
> http://projects.archlinux.org/users/allan/pacman.git/commit/?id=ce40da51611bf6dc236d814894fc90bd624bdc2f

I'm not convinced on this one as it has to scan the entire pkg/
directory, which for some things, could be gigantic (icc,
go-openoffice, etc.). Am I being over-performance crazy? I guess the
zip is going to do it too...

To me this seems way outside the realm of makepkg and belongs in namcap.

-Dan

Re: [PATCH] docs: set text width to 72 in vim directive

2009-12-02T16:47:41Z

On Wed, Dec 2, 2009 at 12:05 PM, Cedric Staniewski <cedric@...> wrote:

A "why" in the commit message would be awfully nice, I'm confused here. :)

> Signed-off-by: Cedric Staniewski <cedric@...>

Re: makepkg in a constrained environment

2009-12-02T16:41:51Z

On Wed, Dec 2, 2009 at 2:47 PM, Ciprian Dorin, Craciun
<ciprian.craciun@...> wrote:

> Looking closer at what the mentioned patch introduced:
> ~~~~
> - local sum="$(${integ}sum "$file" | cut -d ' ' -f 1)"
> + local sum="$(openssl dgst -${integ} "$file" | awk '{print $NF}')"
> ~~~~
> I'm proposing not to get back to what it has been (which assumed
> that there was a md5sum sha128sum, <something-else>sum), but I'm
> proposing that for md5 we should use md5sum, which I'm sure it exists
> on any UNIX. Or better even, try to see if openssl exists, if not
> falling back to md5sum, at least for md5sums...

As a side note, it doesn't exist on all UNIX-y systems, for example on
Mac OS X- there is a /sbin/md5, but not an md5sum command.

-Dan

Re: inside PKGBUILD customizable makepkg PURGE_TARGETS

2009-12-02T16:39:29Z

On Wed, Dec 2, 2009 at 3:48 PM, Allan McRae <allan@...> wrote:

This seems like complete overkill for makepkg to support. I'm -1 on this.

If upstream builds can't get their act together and provide individual
make commands, file a issue upstream. If that doesn't work, then
PURGE_TARGETS is really no more than a

for file in glob_exp; do
rm -f file
done

which is really just a one-liner in a PKGBUILD anyway. Why complicate
the matter?

-Dan

Re: inside PKGBUILD customizable makepkg PURGE_TARGETS

2009-12-02T14:17:23Z

On Wed, Dec 2, 2009 at 11:48 PM, Allan McRae <allan@...> wrote:

> Ciprian Dorin, Craciun wrote:
>>
>> Hello all!
>>
>> Again I have a proposal to enhance makepkg PURGE_TARGETS handling.
>>
>> I think that enabling customizing PURGE_TARGETS inside PKGBUILD
>> would be a benefit in the following context:
>
> Without checking, I would have thought that exporting "PURGE_TRAGETS=..."
> within a package function would just work. I guess the old value would not
> be restored afterwards though.
>
>> * I have a software package that contains: an application, a
>> static library, a dynamic library, and some development files;
>> * I want to create separate packages: something-bin, something-lib
>> (shared library), something-dev (include and static library), and
>> maybe something-doc;
>>
>> What I have to do today (maybe there is a different way, please
>> correct me): I create a `something-lib_package` function in which I
>> make a normal install, but then I delete all unneeded files.
>
> Given this is a fairly common thing to do in the Linux world, most software
> supports this in the Makefiles. "make install-libs" "make install-docs" etc.

Unfortunately not all open source software behaves as nice as
this... (Actually my experience is on the contrary.)
:) As trivia, even pacman doesn't support install-docs, or
install-libs target.

So a simple solution to this problem, that can be applied as a
recipe for most packages, would be: just let the package install as it
wants, and then we'll trim off what we don't want.

Ciprian.

Re: inside PKGBUILD customizable makepkg PURGE_TARGETS

2009-12-02T13:48:05Z

Ciprian Dorin, Craciun wrote:
> Hello all!
>
> Again I have a proposal to enhance makepkg PURGE_TARGETS handling.
>
> I think that enabling customizing PURGE_TARGETS inside PKGBUILD
> would be a benefit in the following context:

Without checking, I would have thought that exporting
"PURGE_TRAGETS=..." within a package function would just work. I guess
the old value would not be restored afterwards though.

> * I have a software package that contains: an application, a
> static library, a dynamic library, and some development files;
> * I want to create separate packages: something-bin, something-lib
> (shared library), something-dev (include and static library), and
> maybe something-doc;
>
> What I have to do today (maybe there is a different way, please
> correct me): I create a `something-lib_package` function in which I
> make a normal install, but then I delete all unneeded files.

Given this is a fairly common thing to do in the Linux world, most
software supports this in the Makefiles. "make install-libs" "make
install-docs" etc.

Re: makepkg in a constrained environment

2009-12-02T13:30:36Z

On Wed, Dec 2, 2009 at 11:16 PM, Xavier <shiningxc@...> wrote:

> On Wed, Dec 2, 2009 at 3:58 PM, Cedric Staniewski <cedric@...> wrote:
>>
>> I do not agree with your definition of "portable" here. All of makepkgs
>> dependencies are available for (mostly?) every system pacman can run on,
>> so it is already "portable". Adding fallbacks only may make it easier to
>> install/run makepkg in a specific setup.
>> A setup where you cannot install all of makepkg's dependencies with a
>> reasonable effort seems rather exotic to me and does not justify such an
>> inappropriate increase in complexity of makepkg in my opinion.
>> Why do you not write bash wrappers which provide the functionality of
>> bsdtar/openssl/... and add these to your path?
>>
>>
>
> I often agree with Cedric.. this is just another occurence :)

:) Ok. Got it: makepkg should make all assumptions possible that
it's working on a fat Linux...

For the moment, I think that my best option is to write my own makepkg.

Of course I'll write it in any other programming language than
Bash... :) For example I think I'll try rc (Plan9 shell), because it
seems the most sane shell I've ever seen...

Thanks,
Ciprian.

Re: makepkg in a constrained environment

2009-12-02T13:16:13Z

On Wed, Dec 2, 2009 at 3:58 PM, Cedric Staniewski <cedric@...> wrote:

>
> I do not agree with your definition of "portable" here. All of makepkgs
> dependencies are available for (mostly?) every system pacman can run on,
> so it is already "portable". Adding fallbacks only may make it easier to
> install/run makepkg in a specific setup.
> A setup where you cannot install all of makepkg's dependencies with a
> reasonable effort seems rather exotic to me and does not justify such an
> inappropriate increase in complexity of makepkg in my opinion.
> Why do you not write bash wrappers which provide the functionality of
> bsdtar/openssl/... and add these to your path?
>
>

I often agree with Cedric.. this is just another occurence :)